Transplantation may be considered for patients with advanced cirrhosis, however, adequate criteria for evaluating survival in those patients are ill-defined. The aim of the present study was to select, among several clinical and functional variables those that could best predict survival at one year. The data collected from 91 consecutive patients with parenchymal cirrhosis hospitalized in our center from February 1984 to January 1986 were subjected to stepwise logistic regression analysis. Death occurring during the first year following entry into the study was considered as a failure. During that period, there were no censored patients. Of 19 variables that entered into the analysis, only two were significant (P less than 0.01): presence (1: moderate; 2: severe) or absence (0) of ascites (A) and breath test (BT: % aminopyrine activity of administered dose at 2 hr). The logistic equation was: 1n (P/1 - P): - 1.95 A + 1.64 BT - 0.393, where P represented the probability of survival at one year. For each patient, P was calculated according to his A and BT values. Using a 0.7 probability cut-point to separate success from failure, 93% (70/75) of successes, 81% (13/16) of failures, and 91% (83/91) of both successes and failures could be correctly predicted. Predictive equations like the present preliminary one can be used in the future to better assess the risk of mortality in patients with parenchymal cirrhosis in whom liver transplantation is considered.